FoodyFocus

Restaurants need good photographs for delivery apps and menus, and a professional shoot costs 400 to 800 euros. FoodyFocus lets an owner photograph the dish on a phone and get a professional image back in seconds. The audience does not buy AI; it buys photographs, and every decision on the site follows from that.

Before any redesign I audited what a crawler actually receives. The site had 334 URLs in five languages and roughly three organic visits a month against a competitor's 1,352. That is a 450x gap, and it was a topic-selection gap rather than a page-count gap.

Two things were holding the site back. About 56 landing pages had never been crawled because the sitemap had grown since Google last read it, which is a twenty-minute dashboard fix that gates everything downstream. And every marketing page shipped 1.5 MB of JavaScript, including the entire 826 KB application bundle and the auth client, so a visitor reading a blog post was booting a logged-in app.

The first audit pass also flagged a canonical decision as a bug. It was not; it was a deliberate response to a real duplicate-content signal, and the audit records the retraction so the next person to crawl the site does not repeat it.

• Stop the bleeding first: the one real 404, the Search Console tasks only a human can do.
• Decouple the marketing bundle from the app before splitting anything. Most of the performance win lands here and it pays for itself even if the split is cancelled.
• Then split: marketing on www, product on app, two builds, two deployments.
• Brand before visuals. Four mood boards, deliberately far apart, each stating the claim it makes and what that claim costs. The point is not to pick a pretty one but to decide which claim the brand makes.

It is the clearest example of how I work: research the real problem, decide in writing, then build it myself. The same six-phase process runs on client work; here I also own the consequences.